Vikram,

The resolver in question has an accuracy of 3 minutes, or 1/20 of a degree. Over 360 degrees roatation, that's an accuracy of 1 part in 7200, or 13 bits accuracy, not resolution. I think you'll want more resolution than 13 bits so you can use several samples to filter noise, predict velocity, etc.

You've correctly assessed the need to excite the resolver with ac. If you want to find the angle from the ratio of the sine and cosine windings, it can be argued you needn't correct for slight variation in the excitation voltage since it cancels in computation of the ratio. However, at angles near zero, 90, 180 and 270 degrees one of the tangent components becomes close to zero resulting a ratio of large and small values - maybe exceeding the acceptable error from a computational algorithm or a lookup table.

Based on the potential for inaccuracies above, use of a demodulator of synchronous or asynchronous type to detect the sine a cosine values may not be the most desireable method of measuring the resolver shaft angle.

Another method is to use a weighted A/D converter in an electronic servo loop to track the resolver output. If the servo loop uses an analog nulling loop with two integrators, the resolver velocity can be estimated with a theoretical error of zero should that be important to you.

Read the data sheets on Analog Devices website (www.analogdevices.com) to get a sense of how this method works. Take a look at the ADS280A and the ADS1200A device datatsheets.

After reading the datasheets, determine whether a software implementation of the converter function is appropriate, or maybe a purchased solution (the converter) is appropriate.

Cheers,

Bruce